Contents lists available at ScienceDirect Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed A stochastic network-based model to simulate the spread of pancreas disease (PD) in the Norwegian salmon industry based on the observed vessel movements and seaway distance between marine farms Sara Amirpour Haredasht a , Saraya Tavornpanich b , Mona Dverdal Jansen b , Trude Marie Lyngstad b , Tadaishi Yatabe a , Edgar Brun b , Beatriz Martínez-López a, ⁎ a Center for Animal Disease Modeling and Surveillance (CADMS), Department of Medicine & Epidemiology, School Veterinary Medicine, University of California, Davis, CA, USA b Norwegian Veterinary Institute, Oslo, Norway ARTICLE INFO Keywords: Aquaculture Data-based model Spatio-temporal dynamics Risk-based surveillance Epidemiology ABSTRACT Pancreas disease (PD) is a viral disease of economic importance affecting farmed Atlantic salmon (Salmo salar L.) and rainbow trout (Oncorhyncus mykiss (Walbaum)) in the seawater phase in Ireland, Norway and Scotland. In this study we used a stochastic network-based disease spread model to better understand the role of vessel movements and nearby seaway distance on PD spread in marine farms. We used five different edge’s definitions and weights for the network construction: high-risk vessel movements, high-risk wellboat movements and high- risk nearby seaway distance at < 20 km, < 10 km or < 5 km, respectively. Models were used to simulate PD spread in marine farms as well as to simulate the spread of marine SAV2 and SAV3 subtypes independently and results were compared with the observed PD, marine SAV2 and SAV3 cases in Norway in 2016. Results revealed that the model that provided the best fit of the observed data and, therefore, the one considered more biolo- gically plausible, was the one using high-risk wellboat movements. The marine SAV2, SAV3 and PD models using wellboat movements were able to correctly simulate the farms status (PD positive or PD negative) with the sensitivity of 84%, 85%, 84% and Specificity of 98%, 97% and 94%, respectively. These results should contribute to inform more cost-effective prevention and control policies to mitigate PD spread and to improve the sus- tainability and long-term profitability of the salmon industry in Norway. 1. Introduction Pancreas disease (PD) is a viral disease of economic importance affecting farmed Atlantic salmon (Salmo salar L.) and rainbow trout (Oncorhyncus mykiss (Walbaum)) in the seawater phase in Ireland, Norway and Scotland (Jansen et al., 2017). PD is a reportable disease in Norway and the infection with the causative agent of PD has been in- cluded to the World Organisation for Animal Health (OIE) list of fish diseases since 2014 (OIE, 2016). The causal agent, Salmonid alphavirus (SAV), can be divided into six subtypes (subtypes 1–6, Fringuelli et al., 2008). Norway currently has two endemic SAV subtypes: SAV3 has only been identified in Norway, and marine SAV2 has been described in both Norway and Scotland (Fringuelli et al., 2008; Hjortaas et al., 2016). The presence of these two subtypes is largely separated into two endemic areas along the coast, SAV3 in South West part of Norway, south of Møre og Romsdal county, and marine SAV2 in Møre og Romsdal and Trøndelag counties, with some degree of overlap at the border area (Jansen et al., 2017). In recent years, the number of new PD cases in Norway has been stable at around 140 cases per year, with the pro- portion of SAV3 to marine SAV2 cases varying a little from year to year (Hjeltnes et al., 2017). The main route of PD transmission has been suggested to be the horizontal spread between seawater sites from in- fected farms to nearby farms (Viljugrein et al., 2009; Kristoffersen et al., 2009; Aldrin et al., 2010; Tavornpanich et al., 2012; Stene et al., 2014), however the duration of SAV-survival in water and the specific water conditions (e.g., temperature, currents) affecting such survival and transmission mechanism(s) is unknown. Reduced biosecurity, through sharing of equipment and personnel, has also been shown to increase the risk of PD (McLoughlin et al., 2003; Kristoffersen et al., 2009; Aldrin et al., 2010; Stene et al., 2014). Information on wellboat activity in Norwegian aquaculture, including information about the fleet, regula- tions, transport activity and types of transports (for example movement https://doi.org/10.1016/j.prevetmed.2018.05.019 Received 9 June 2017; Received in revised form 8 January 2018; Accepted 31 May 2018 ⁎ Corresponding author. E-mail address: beamartinezlopez@ucdavis.edu (B. Martínez-López). Preventive Veterinary Medicine xxx (xxxx) xxx–xxx 0167-5877/ © 2018 Published by Elsevier B.V. Please cite this article as: Amirpour Haredasht, S., Preventive Veterinary Medicine (2018), https://doi.org/10.1016/j.prevetmed.2018.05.019